Nature scenario plausibility: A dynamic Bayesian network approach
Chiara Colesanti Senni, Skand Goel · Ecological Economics · 2025
To cope with the lack of quantifiable knowledge about the occurrence of nature-related risks, scenario analysis has emerged as a way to investigate possible futures. We argue that expressing scenario narratives as causal models – leveraging causal Bayesian graphs – opens up new avenues for designing and using scenarios. As one use case of this approach, we show how dynamic Bayesian networks to assess the plausibility of high-dimensional quantitative scenarios. We provide an algorithm that probabilistically evaluates whether a quantitative scenario is consistent with a certain narrative about nature-economy linkages. This can allow the user to choose among several available scenarios using a data-driven approach. As a demonstration, we apply this approach to data from an integrated assessment model.